[1.26.05] Fix count of upper-case letters in a word to consider multibyte characters.

Mbrola translations, option to not match across a word boundary.
Mbrola translations, allow check of lengthen indicator [:].
espeakedit, add operation to apply a voice's formants modifications to a keyframe.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@47 d46cf337-b52f-0410-862d-fd96e6ae7743

src/menus.cpp

@@ -129,7 +129,7 @@ wxMenuBar *MakeMenu(int type)
 	menu_bar->Append(voice_menu, _T("&Voice"));
 	menu_bar->Append(option_menu, _T("&Options"));
 	menu_bar->Append(tools_menu, _T("&Tools"));
-	menu_bar->Append(data_menu, _T("&Data"));
+	menu_bar->Append(data_menu, _T("&Compile"));
 	menu_bar->Append(help_menu, _T("&Help"));
 
 	return(menu_bar);

src/readclause.cpp

@@ -239,8 +239,11 @@ static int GetC(void)
 	int c;
 	int c1;
 	int c2;
+	int cbuf[4];
+	int ix;
 	int n_bytes;
 	unsigned char m;
+	static int ungot2 = 0;
 	static const unsigned char mask[4] = {0xff,0x1f,0x0f,0x07};
 	static const unsigned char mask2[4] = {0,0x80,0x20,0x30};
 
@@ -250,7 +253,16 @@ static int GetC(void)
 		return(c1);
 	}
 
-	c1 = GetC_get();
+	if(ungot2 != 0)
+	{
+		c1 = ungot2;
+		ungot2 = 0;
+	}
+	else
+	{
+		c1 = GetC_get();
+	}
+
 	if(option_multibyte == espeakCHARS_WCHAR)
 	{
 		count_characters++;
@@ -271,14 +283,13 @@ static int GetC(void)
 		if(((c1 & 0xf8) == 0xf0) && ((c1 & 0x0f) <= 4))
 			n_bytes = 3;
 
-		if(n_bytes > 0)
+		if((ix = n_bytes) > 0)
 		{
-			c = c1 & mask[n_bytes];
-			m = mask2[n_bytes];
-			while(n_bytes > 0)
+			c = c1 & mask[ix];
+			m = mask2[ix];
+			while(ix > 0)
 			{
-				c2 = GetC_get();
-				if(c2 == 0)
+				if((c2 = cbuf[ix] = GetC_get()) == 0)
 				{
 					if(option_multibyte==espeakCHARS_AUTO)
 						option_multibyte=espeakCHARS_8BIT;   // change "auto" option to "no"
@@ -288,14 +299,17 @@ static int GetC(void)
 
 				if((c2 & 0xc0) != 0x80)
 				{
+					// This is not UTF8.  Change to 8-bit characterset.
+					if(n_bytes > 1)
+						ungot2 = cbuf[2];
 					GetC_unget(c2);
 					break;
 				}
 				m = 0x80;
 				c = (c << 6) + (c2 & 0x3f);
-				n_bytes--;
+				ix--;
 			}
-			if(n_bytes == 0)
+			if(ix == 0)
 			{
 				count_characters++;
 				return(c);

src/spect.cpp

@@ -377,6 +377,21 @@ void SpectFrame::ToggleMarker(int n)
 }
 
 
+void SpectFrame::ApplyVoiceMods()
+{//==============================
+	// apply the modifications to the formants which are defined in the current voice
+	int pk;
+
+	for(pk=0; pk<N_PEAKS; pk++)
+	{
+		peaks[pk].pkfreq = (peaks[pk].pkfreq * voice->freq2[pk])/256;
+		peaks[pk].pkheight = (peaks[pk].pkheight * voice->height2[pk])/256;
+		peaks[pk].pkwidth = (peaks[pk].pkwidth * voice->width2[pk])/256;
+		peaks[pk].pkright = (peaks[pk].pkright * voice->width2[pk])/256;
+	}
+}
+
+
 
 double SpectFrame::GetRms(int seq_amplitude)
 {//=========================================

src/spect.h

@@ -98,6 +98,7 @@ public:
 	void ZeroPeaks();
 	void CopyPeaks(SpectFrame *sf);
 	void ToggleMarker(int n);
+	void ApplyVoiceMods();
 	
 	void MakeWave(int peaks, PitchEnvelope &pitch, int amplitude, int duration);
 	void MakeHtab(int numh, int *htab, int pitch);

src/spectdisplay.cpp

@@ -468,7 +468,7 @@ void SpectDisplay::OnMenu(wxCommandEvent& event)
 	int code;
 	wxKeyEvent keyevent;
 	SpectFrame *sf;
-	static int key[] = {0x1044,0x3044,0x104e,0x1047,0x105a,0x1051,0x3051,WXK_F1,WXK_F2,0x1049};
+	static int key[] = {0x1044,0x3044,0x104e,0x1047,0x105a,0x1051,0x3051,WXK_F1,WXK_F2,0x1049,WXK_F10};
 
 	id = event.GetId();
 
@@ -773,6 +773,11 @@ void SpectDisplay::OnKey(wxKeyEvent& event)
 		display = 2;
 		break;
 
+	case WXK_F10:
+		sf->ApplyVoiceMods();
+		display = 2;
+		break;
+
 	case WXK_F2:
 		// make and play sequence
 		ReadDialogValues();
@@ -1002,6 +1007,7 @@ void InitSpectrumDisplay()
 	menu_spectdisplay->Append(207,_T("Hide Interpolation	CTRL+SHIFT+Q"));
 	menu_spectdisplay->Append(205,_T("Zero Peaks	CTRL+Z"));
 	menu_spectdisplay->Append(210,_T("Toggle Marker"),menu_markers);
+	menu_spectdisplay->Append(211,_T("Apply the formant adjustments from current voice"));
 	menu_spectdisplay->AppendSeparator();
 	menu_spectdisplay->Append(208,_T("Play Selected Frame	F1"));
 	menu_spectdisplay->Append(209,_T("Play All	F2"));

src/synth_mbrola.cpp

@@ -196,6 +196,7 @@ int GetMbrName(PHONEME_LIST *plist, PHONEME_TAB *ph, PHONEME_TAB *ph_prev, PHONE
 	// bit 0  skip the next phoneme
 	// bit 1  match this and Previous phoneme
 	// bit 2  only at the start of a word
+	// bit 4  don't match two phonemes across a word boundary
 
 	pr = mbrola_tab;
 	while(pr->name != 0)
@@ -205,9 +206,17 @@ int GetMbrName(PHONEME_LIST *plist, PHONEME_TAB *ph, PHONEME_TAB *ph_prev, PHONE
 			if(pr->next_phoneme == 0)
 				found = 1;
 			else
+			if((pr->next_phoneme == ':') && (plist->synthflags & SFLAG_LENGTHEN))
+			{
+				found = 1;
+			}
+			else
 			{
 				if(pr->control & 2)
 					other_ph = ph_prev;
+				else
+				if((pr->control & 8) && ((plist+1)->newword))
+					other_ph = phoneme_tab[phPAUSE];  // don't match the next phoneme over a word boundary
 				else
 					other_ph = ph_next;
 

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.26.04  13.Jun.07";
+const char *version_string = "1.26.05  14.Jun.07";
 const int version_phdata  = 0x012601;
 
 int option_device_number = -1;

src/tr_languages.cpp

@@ -271,6 +271,7 @@ Translator *SelectTranslator(const char *name)
 
 
 	case L('h','r'):   // Croatian
+	case L('s','r'):   // Serbian
 		{
 			static const wchar_t replace_chars_hr[] = 
 				{0x430,0x431,0x446,0x45b,0x447,0x434,0x452,0x45f,
@@ -389,6 +390,10 @@ Translator *SelectTranslator(const char *name)
 			tr = new Translator();
 			tr->charset_a0 = charsets[4];   // ISO-8859-4, includes a,e,i,o,u-macron
 			tr->langopts.stress_rule = 2;
+			tr->langopts.stress_flags = 0x21;
+			tr->langopts.unstressed_wd1 = 0;
+			tr->langopts.unstressed_wd2 = 2;
+			tr->langopts.param[LOPT_DIERESES] = 1;
 		}
 		break;
 

src/translate.cpp

@@ -1601,6 +1601,7 @@ void *Translator::TranslateClause(FILE *f_text, const void *vp_input, int *tone_
 	int dict_flags;        // returned from dictionary lookup
 	int word_flags;        // set here
 	int embedded_count = 0;
+	int letter_count = 0;
 	char *word;
 	char *p;
 	int j, k;
@@ -1854,6 +1855,8 @@ if((c == '/') && (langopts.testing & 2) && isdigit(next_in) && IsAlpha(prev_out)
 			{
 				if(!IsAlpha(prev_out) || (langopts.ideographs && (c >= 0x3000)))
 				{
+					letter_count = 0;
+
 					if((prev_out != ' ') && (prev_out != '\''))
 					{
 						// start of word, insert space if not one there already
@@ -1875,6 +1878,8 @@ if((c == '/') && (langopts.testing & 2) && isdigit(next_in) && IsAlpha(prev_out)
 					}
 				}
 
+				letter_count++;
+
 				if(iswupper(c))
 				{
 					c = towlower(c);
@@ -1909,7 +1914,7 @@ if((c == '/') && (langopts.testing & 2) && isdigit(next_in) && IsAlpha(prev_out)
 						source_index = prev_source_index;  // unget
 					}
 #endif
-					if((all_upper_case) && ((ix - words[word_count].start) > 1))
+					if((all_upper_case) && (letter_count > 2))
 					{
 						if((c == 's') && (next_in==' '))
 						{